Table 2. Predictive power, validation, and interactions.
| Model | AUC | MSE1 | Misclassification |
|---|---|---|---|
| REGARDS: Development2 | 0.789 | 0.099 | 0.131 |
| ARIC: External validation | 0.846 | 0.074 | 0.090 |
| Interactions3 | |||
| Sex and race with main effects4 | 0.794 | 0.098 | 0.130 |
| SBP × DBP5 | 0.788 | 0.099 | 0.131 |
| MAP6 | 0.789 | 0.099 | 0.131 |
| HDL × triglycerides7 | 0.779 | 0.100 | 0.133 |
| Waist circumference × BMI8 | 0.780 | 0.100 | 0.132 |
| BMI × HDL9 | 0.787 | 0.099 | 0.131 |
| BMI × triglycerides10 | 0.785 | 0.100 | 0.132 |
1MAP calculated as the average squared difference between the observed and predicted values.
2Analyzed using Bayesian logistic regression. Diabetes incidence ~ age + sex + race + BMI + triglycerides + HDL cholesterol + SBP + DBP + blood glucose.
3Analyzed using the REGARDS dataset by Bayesian logistic regression.
4Diabetes incidence ~ (age + BMI + triglycerides + HDL cholesterol + SBP + DBP + blood glucose) × (sex:race).
5Diabetes incidence ~ age + sex + race + BMI + triglycerides + HDL cholesterol + SBP:DBP + blood glucose.
6MAP calculated as [(2 ×DBP) + SBP]/3. Diabetes incidence ~ age + sex + race + BMI + triglycerides + HDL cholesterol + MAP + blood glucose.
7Diabetes incidence ~ age + sex + race + BMI + triglycerides:HDL cholesterol + SBP + DBP + blood glucose.
8Diabetes incidence ~ age + sex + race + BMI:waist circumference + triglycerides + HDL cholesterol + SBP + DBP + blood glucose.
9Diabetes incidence ~ age + sex + race + BMI:HDL cholesterol + triglycerides + SBP + DBP + blood glucose.
10Diabetes incidence ~ age + sex + race + BMI:triglycerides + HDL cholesterol + SBP + DBP + blood glucose.
ARIC, Atherosclerotic Risk in Communities; AUC, area under the curve; DBP, diastolic blood pressure; HDL, high-density lipoprotein; MAP, mean arterial pressure; MSE, mean squared error; REGARDS, REasons for Geographic And Racial Differences in Stroke; SBP, systolic blood pressure.